Positive regulatory effects of perioperative probiotictreatment on postoperative liver complications aftercolorectal liver metastases surgery: a double-centerand double-blind randomized clinical trialLiu et al.

Liu et al. BMC Gastroenterology (2015) 15:34 DOI 10.1186/s12876-015-0260-z

Liu et al. BMC Gastroenterology (2015) 15:34 DOI 10.1186/s12876-015-0260-z

RESEARCH ARTICLE Open Access

Positive regulatory effects of perioperativeprobiotic treatment on postoperative livercomplications after colorectal liver metastasessurgery: a double-center and double-blindrandomized clinical trialZhihua Liu1,2*, Chao Li1, Meijin Huang1, Chao Tong1, Xingwei Zhang1, Lei Wang1, Hui Peng1, Ping Lan1,Peng Zhang2, Nanqi Huang1, Junsheng Peng1, Xiaojian Wu1, Yanxing Luo1, Huanlong Qin2*, Liang Kang1*

and Jianping Wang1*

Abstract

Background: Colorectal liver metastases (CLM) occur frequently and postoperative intestinal infection is a commoncomplication. Our previous study showed that probiotics could decrease the rate of infectious complications aftercolectomy for colorectal cancer. To determine the effects of the perioperative administration of probiotics on serumzonulin levels which is a marker of intestinal permeability and the subsequent impact on postoperative infectiouscomplications in patients with CLM.

Methods: 150 patients with CLM were randomly divided into control group (n = 68) and probiotics group (n = 66).Probiotics and placebo were given orally for 6 days preoperatively and 10 days postoperatively to control groupand probiotics group respectively. We used the local resection for metastatic tumor ,while for large tumor, thesegmental hepatectomy. Postoperative outcome were recorded. Furthermore, complications in patients withnormal intestinal barrier function and the relation with serum zonulin were analyzed to evaluate the impact on theliver barrier dysfunction.

Results: The incidence of infectious complications in the probiotics group was lower than control group. Analysisof CLM patients with normal postoperative intestinal barrier function paralleled with the serum zonulin level. Andprobiotics could also reduce the concentration of serum zonulin (P = 0.004) and plasma endotoxin (P < 0.001).

Conclusion: Perioperative probiotics treatment could reduce the serum zonulin level, the rate of postoperativesepticemia and maintain the liver barrier in patients undergoing CLM surgery. we propose a new model about theregulation of probiotics to liver barrier via clinical regulatory pathway. We recommend the preoperative oral intakeof probiotics combined with postoperative continued probiotics treatment in patients who undergo CLM surgery.

Trial registration: ChiCTR-TRC-12002841. 2012/12/21

Keywords: Probiotics, Colorectal liver metastases, Zonulin, Postoperative complication, Liver barrier

* Correspondence: liuzhh26@mail.sysu.edu.cn; hlqin@live.cn;lkang1972@hotmail.com; j-p-wang@hotmail.com1Gastrointestinal Institute of Sun Yat-sen University, Department of ColorectalSurgery, the Sixth Affiliated Hospital of Sun Yat-sen University (GuangdongGastrointestinal Hospital), 26 Yuancun Erheng Road, Guangzhou, Guangdong510655, People’s Republic of China2Department of Surgery, Shanghai JiaoTong University Affiliated SixthPeople’s Hospital, Shanghai 200233, China

© 2015 Liu et al.; licensee BioMed Central. ThiAttribution License (http://creativecommons.oreproduction in any medium, provided the orDedication waiver (http://creativecommons.orunless otherwise stated.

s is an Open Access article distributed under the terms of the Creative Commonsrg/licenses/by/4.0), which permits unrestricted use, distribution, andiginal work is properly credited. The Creative Commons Public Domaing/publicdomain/zero/1.0/) applies to the data made available in this article,

Liu et al. BMC Gastroenterology (2015) 15:34 Page 2 of 13

BackgroundThe liver is a common site for metastatic disease [1].Over 50% of the primary tumor are originated from thegastrointestinal tract, especially colon and rectum [2].Because morbidity and mortality of colorectal cancer(CRC) increased year by year, CRC has been ranked asthe second cause of tumor death around the world [3]and liver is the most common site of metastases of CRC[4-6]. Resection of colorectal liver metastases (CLM) isthe best choice for the treatment of CRC with liver metas-tasis [7], it is reported that simultaneous resection is as ef-ficient as a delayed procedure for long-term survival [8].With the development of advanced surgical tech-

niques, postoperative survival rate of CLM has beenconsiderably promoted [9]. However, the incidence ofpostoperative infectious complications are becomingmore and more common [10,11]. These complicationsmay also lead to higher rates of recurrence and death[12-15].It is known to all that PRO play an important role in

the stability of the intestinal microbiological environment[16,17]. In recent years, some randomized clinical trialsusing PRO for to surgical patients for the protectiveeffects perioperatively [18-20] could regulate intestinalmicrobial populations, lower intestinal permeability,decrease the incidence of infection-related complica-tions and overcome other problems, significantly.Zonulin, the only known physiological modulator of

intercellular TJ described so far, is a protein which con-stitutes tight junctions of the digestive tract and modu-lates intestinal permeability [1,21]. Recently, zonulin wasfound to be a reflection of the intestinal permeability[22]. It is reported that elevated levels of plasma zonulinin septic patients might serve as a mechanism forincreased intestinal permeability in sepsis and systemicinflammatory response syndrome (SIRS) [23]. Anotherstudy showed that zonulin both in vitro and in vivoinduced generation of complement C3a and C5a, sug-gesting that zonulin facilitated the development of acutelung injury and lung barrier by enhancing albumin leakand complement activation as well as increased build upof neutrophils and cytokines [24]. However, no study hasexplored the link about zonulin, liver barrier and therates of complications after CLM surgery, the specificrole of PRO has not been elucidated yet.Our previous study indicated that PRO could increase

transepithelial resistance (TER) and increase fecal bac-terial variety, reduce transmucosal permeability of horse-radish peroxidase, lactulose/mannitol ratio, bacterialtranslocation rate, ileal-bile acid binding protein, positiverate of blood bacterial DNA, blood enteropathogenicbacteria [11]. Further study showed that PRO decreasedthe serum zonulin concentration, the duration of post-operative pyrexia, duration of antibiotic therapy and the

rate of postoperative infective complications in patientsunderwent colectomy [1].As liver barrier may have an impact on the incidence

of postoperative liver complications, we proposed thehypothesis that pre- and postoperative PRO may reduceliver permeability, decrease the rate of bacterial trans-location (BT) and infectious complications after CLMsurgery. we aimed to investigate the effects of the peri-operative administration of PRO on serum zonulinlevels, liver permeability and the subsequent impact onpostoperative liver complications in patients undergoingCLM.

MethodsInclusion and exclusion criteriaInclusion criteria in our study included: 1) age between25 and 75 years; 2) the diagnosis of CRC were confirmedby biopsy and histological testing, liver metastases werediagnosed by CT preoperatively and confirmed by post-operative histological report; 3) patients underwent aradical resection of primary colorectal tumors and livermetastases; and 4) they had no other metastasis.The exclusion criteria were as follows: 1) pregnant; 3)

lactose intolerance; 4) clinically significant immunodefi-ciency; 5) gastrointestinal disorders (e.g., Crohn’s diseaseor ulcerative colitis); 6) received antibiotics during the10 days before surgery; 7) infection; 8) probiotic or ex-cessive fiber intake within 2 weeks of surgery; 9)under-went an emergency operation; 10) bowel preparation forcolonoscopy prior to surgery within 6 days; 11)under-went a proctectomy with low rectal anastomosis orsurgery for a polypoid lesion; 12) the surgery was per-formed laparoscopically; 13) received preoperative neoad-juvant chemotherapy or radiotherapy; 14) had otherdistant metastases except liver; 15) unresectable liver me-tastases; and 16) severe liver function failure or otherorgan failure.

Patients150 patients with CLM were scheduled to undergo col-ectomy at the Shanghai Sixth People’s Hospital, affiliatedto Shanghai JiaoTong University in Shanghai or theSixth Affiliated Hospital of Sun Yat-sen University inGuangzhou, between April 2007 and July 2013. The pa-tients were randomized prior to surgery to the placebocontrol group (control) with perioperative oral feedingplacebo or the PRO therapy group (PRO) with PROtreatment pre- and post-operatively. The study protocolswere reviewed and approved by the Human ResearchReview Committee of the two hospitals and written in-formed consent for participation were obtained fromeach patient [1]. The requirement for informed consentwas waived by each of the institutional review boards(IRBs) that approved the study (Human Research Review

Liu et al. BMC Gastroenterology (2015) 15:34 Page 3 of 13

Committee of the The Sixth Affiliated Hospital of SunYat-sen University).All patients were assessed for eligibility, while 16 pa-

tients were excluded, for whom did not meet the inclu-sion criteria (10 patients) or refused to participate (6patients). 117 of the 134 patients completed the entirestudy (Figure 1). There were no significant differences insex, age, BMI, time between the onset of symptoms andhospital admission, no significant difference was foundin the preoperative serum levels of albumin, Hb, creatin-ine, and operative time, intra-operative blood loss, intra-operative transfusion, usage of supplemental albuminpostoperation, preoperative preparation time the num-ber of patients treated with metronidazole, penicillin,ceftriaxone, liver function (ALT and AST) (Table 1 forintention-to-treat and Additional file 1: Table S1 for per-protocol).

Study design, PRO treatment, and patient careRandomization to two groups was accomplished usingdouble randomization principle (with the help of enve-lopes and random variation row) constitution in our pa-tient department of the Sixth Affiliated Hospital of SunYat-sen University. Only a nurse knew the treatment as-signment who was not directly involved in the trial andcould broke the treatment codes in the event of anemergency.Patients in the PRO group received encapsulated ad-

mixture of three PRO bacteria (Institute of LifeScience of Only, Shanghai Jiao Tong University,Shanghai, China), composed of LP (CGMCC No.1258,cell count ≥1011 cfu/g), LA-11 (cell count ≥7.0 ×1010 cfu/g) and BL-88 (cell count ≥5.0 × 1010 cfu/g)

Figure 1 Flow chart of the randomization procedure used to enroll p

every day. An acid-resistant coating was used to pre-pare the capsules which wrapping the PRO a or pla-cebo. Each patient in the PRO group received PRO2 g/day, at a total daily dose of 2.6 × 1014 cfu. Patientsin the control group received daily encapsulatedmaltodextrin. The appearance, smell and taste of thetwo types of capsules showed no obvious difference.The intervention period lasted 16 days, 6 days pre-

operatively and 10 days postoperatively. All the subjectswere interviewed by the study nurse, any reactions tothe product, medications taken and adverse events thatoccurred were recorded.During the study period, no parenteral or enteral nu-

tritional supplementation was given. All patients re-ceived a regular diet preoperatively, and a low-residuediet 1 day preoperatively. Mechanical bowel preparationwas given 1 day before the surgery, in which all patientswere given Soffodex, containing 2.4 g of monobasic so-dium phosphate and 0.9 g of dibasic sodium phosphate.Parenteral hydration was given on the morning of thesurgery via a central venous catheter. A catheter wasplaced for gastric aspiration to reduce flow through thecolon anastomosis. 500 mg of metronidazole and 1 g ofceftriaxone were given 1 h before induction and contin-ued for 48 h after surgery. After the surgery, all patientsreceived regular parenteral hydration. Complicationswere registered daily and patients were re-examined atthe outpatient clinic 1, 2, and 4 weeks after surgery.

Postoperative clinical observationsDetailed postoperative records were kept daily and infec-tious complications were recorded for up to 30 daysafter surgery. The diagnosis of bacterial infection was

atients in the study.

Table 1 Baseline of characteristics of the patients with colorectal liver metastases undergoing surgery at hospitaladmission in the study (Intention-to-treat)

Index Control group (n = 68) PRO group (n = 66)

Sex (Male/Female) 35/33 35/31

Age (Year) 60.16 ± 16.20 65.62 ± 18.18

BMI (kg/m2) 23.06 ± 5.26 22.28 ± 3.66

Location of tumor

ascending colon 15 15

Transverse colon 6 5

Descending colon 12 16

Sigmoid colon 15 11

Rectum 20 19

Time between onset of symptoms and hospital admission (d) 50.28 ± 16.62 55.12 ± 18.26

Preoperative albumin (g/dL) 38.26 ± 8.56 36.98 ± 6.96

Preoperative Hb (g/L) 126.86 ± 32.06 116.22 ± 36.68

Creatinine (mg/dL) 1.26 ± 0.28 1.20 ± 0.62

Operative time (min) 186.28 ± 56.36 199.82 ± 55.98

Intra-operative blood loss (ml) 336.26 ± 182.68 352.56 ± 169.26

Transfusion during operation (ml) 308.12 ± 120.66 329.16 ± 130.58

Usage of supplemental albumin postoperation (g) 22.38 ± 16.36 28.12 ± 20.98

Preoperation prepared time (d) 6.12 ± 3.52 6.68 ± 2.98

Metronidazole (n) 68 66

Penicillin (n) 30 33

Ceftriaxone (n) 38 33

ALT (U/L) 35.68 ± 15.26 32.62 ± 18.86

AST (U/L) 29.68 ± 16.56 28.22 ± 18.86

BMI, body mass index; Hb, hemoglobin; ALT, alanine transarninase (normal value, 0–40 U/L); AST, aspartate aminotransferase (normal value, 0–40 U/L).There were no significant differences about the characteristics between the two groups.Quantitative data are expressed as mean ± standard deviation. Numerical data were compared by t test and nominal data by Pearson χ2 test or Fisher’s exact testbetween groups.

Liu et al. BMC Gastroenterology (2015) 15:34 Page 4 of 13

based on the previous reference [11,19] and several typesof complications were observed after surgery. The septi-cemia incidence, liver function after 10 days postoperative,the use of PRO, intra-abdominal drainage time andinfections-related complication were recorded. One of thethree doctors examine the patients (Liang Kang, PengZhang, and Chao Li) to evaluate bowel sounds, abdominalcramping, and distension on the postoperative days.Our previous study [1] indicated that PRO lower the

intestinal permeability and the postoperative septicemiarate. In this study, we further investigate the effects ofPRO on the liver barrier by comparing the postoperativeinfection-related complications with normal intestinalpermeability at day 3 after surgery to eliminate the inter-ference of intestinal barrier.

Measurement of serum zonulin concentrationThe concentration of serum zonulin collected at 24 hpostoperatively were determined using an ELISA kit[25]. Briefly, plastic microtiter plates (Costar, Cambridge,

MA) were coated with rabbit zonulin cross-reacting Zotderivative ΔG IgG antibodies (10 μg/ml in 0.1 mol/l so-dium carbonate buffer, pH 9.0) [25]. After an overnightincubation at 4°C, plates were washed four times inTBS-T and blocked by incubation for 1 h at 37°C withTBS-T. After four washes in TBS-T, five ΔG serial stan-dards (50, 25, 12.5, 6.2, 3.1, and 0 ng/ml) and patientsera samples (1:101 dilution in TBS-T) were added andincubated overnight at 4°C. After four washes with Tris-buffered saline 0.2% Tween 20 buffer, plates were incu-bated with biotinylated anti-Zot IgG antibodies for 4 hat 4°C. A color reaction was developed using a commer-cial kit (ELISA amplification kit; Invitrogen). The ab-sorbance at 495 nm was measured with a microplateauto-reader (Molecular Devices Thermomax MicroplateReader).

Intestinal permeability assayIntestinal permeability was assessed using the L/M testpreoperatively (the morning of the surgery) and on the

Liu et al. BMC Gastroenterology (2015) 15:34 Page 5 of 13

3th and 10th day postoperatively [1,11]. After an over-night fast, all subjects were given the oral test solutioncontaining 10 g of lactulose (Sigma-Aldrich, Tokyo,Japan) and 5 g of mannitol (Sigma-Aldrich) in 60 ml ofphysiological saline. For the next 6 h, the subjects wererested and no food or water was allowed. Complete 6 hurine collections were taken and a further 10-mL urinesample was frozen at −20°C until analysis. Urinary lactu-lose and mannitol concentrations were measured bygas–liquid chromatography.

Measurement of plasma concentrations of endotoxinPlasma samples used for endotoxin measurements werestored in endotoxin-free glass tubes to prevent the lossof endotoxin to plastic tubes walls. All materials used forthe assay were rendered endotoxin-free. Plasma endo-toxin concentrations were measured by a commerciallyavailable kit [Cambrex Limulus Amebocyte Lysate (LAL)kit; Lonza Inc, Walkersville, MD] [26]. LPS from thesample reacts enzymatically with a proenzyme in theLAL reagent that leads to its activation and the produc-tion of a colored peptide from the chromogenic sub-strate reagent over a short incubation period that can beread at 405–410 nm.

Microbiological investigations and PCR assay for bacterialDNA fragmentClinical samples comprising blood (40 mL), central lines(tips), urine (20 mL), and sputum were collected at06:00 [11]. Each clinical sample was taken approximately72 h after the operation, and immediately sent to themicrobiological laboratory. The specimens were culturedunder aerobic condition, microaerophilic condition andanaerobic condition respectively at 35-37°C for 24–48 h.Anaerobic cultivation was performed in anaerobic cham-ber. Sabouraud’s medium (bioMérieux, France) were usedin the Fungal cultures. The biochemical characteristics ofthe cultured strains were investigated using the API and/or ID tests (bioMérieux, France). The remaining 20 mL ofblood was collected in a sterile container containingEDTA. To determine the sensitivity of the PCR detection,serial dilutions of the spiked blood were tested until anegative result. The sensitivity of the test was 10 organ-isms/mL.

RNA extraction, cDNA synthesis, and real-time quantitativePCR (qRT-PCR)Our previous study indicated that the p38 mitogen-activated protein kinase (MAPK) signaling pathway wasinvolved in the protective process of intestinal perme-ability and intestinal barrier function [27]. Thus, we de-termined the liver p38 MAPK gene expression using theRT-PCR [25,27]. Briefly, total RNA was extracted fromthe adjacent normal liver tissues of five patients with

CLM using Trizol reagent (Invitrogen, Grand Island,NY) to detect the expression level of p38 MAPK. Foreach sample, 600 ng of mRNA was used in the reversetranscription reaction (iScript kit from BioRad Labora-tories, Hercules, CA, USA). Further analysis of themRNA levels of each group was performed by real timePCR with a light-cycling system (LightCycler; RocheDiagnostics GmbH, Mannheim, Germany). All valueswere expressed as a fold increase or decrease comparewith the expression of actin. The sequences of theprimers were as follows: p38 MAPK, F: 5′-GAA-GAGCCTGACCTACGAT-3′ and R: 5′-ACTGCCAAG-GAGCATCTA-3′; β-actin, F: 5′-CTCCATCCTGGCCTCGCTGT-3′ and R: 5′-GCTGTCACCTTCACCGTTCC-3′.

Statistical analysisA sample size calculation based on the prevalence of BTdemonstrated that approximately 120 patients would berequired in each group to demonstrate a reduction inBT from 25% to 5% at the 5% significance level with apower of 80% [1]. Results were analyzed using SPSS 13.0version for Windows (SPSS, Chicago, Illinois, USA).Quantitative data are expressed as mean ± standard devi-ation. For postoperative clinical tests, numerical datawere compared by t-tests and nominal data were ana-lyzed between groups with the Pearson χ2 test or Fisher’sexact test. A P-value <0.05 was considered statisticallysignificant. Spearman’s correlation was used to assessthe relationship between zonulin level and outcomeusing SPSS 13.0.

Results and discussionPRO reduced postoperative intestinal infection relatedcomplications and promoted rapid recoveryWe first performed the comparison of overall postopera-tive outcomes between PRO and control groups (Table 2& Additional file 2: Table S2), which verified our previ-ous study [1]. Among the 134 patients, no death casewas found, no patients got complications related leakageof the anastomosis, fistulas, and abdominal hemorrhage,no side effect of probiotic was reported in our study. Re-sults indicated that the incidence of infectious complica-tions in the PRO group was lower than the controlgroup, such as septicemia incidence (59% vs 88%, P =0.008), urinary infection (2% vs 13%, P = 0.017), diarrheaincidence (24% vs 46%, P = 0.012), then decrease theduration of postoperative pyrexia (6.02 ± 1.68 vs 6.98 ±2.22, P = 0.006), cumulative duration of antibiotic ther-apy (6.22 ± 1.96 vs 7.56 ± 2.26, P < 0.001), postoperativehospital stay (11.26 ± 2.52 vs 12.96 ± 3.06, P < 0.001), andthe overall hospital charge (52261.16 ± 12168.28 vs58262.36 ± 10262.36, P = 0.002). For the liver functionindexes, both ALT and AST were significantly lowered

Table 2 Comparison of postoperative outcomes between probiotics and control (Intention-to-treat)

Outcomes Intention-to-treat

Control (n = 68) PRO (n = 66) P Value

Septicemia incidence (%) 88 (60/68) 59 (39/66) 0.008

ALT (U/L) 56.20 ± 18.16 36.28 ± 18.92 <0.001

AST (U/L) 45.62 ± 22.68 36.18 ± 21.52 0.015

Intro-abdominal drainage time (d) 4.2 ± 1.6 4.6 ± 1.8 0.176

Incision infection (%) 12 (8/68) 9 (6/66) 0.779

Central lines infection (%) 9 (6/68) 11 (7/66) 0.777

Pneumonia infection (%) 12 (8/68) 9 (6/66) 0.097

Urinary infection (%) 13 (9/68) 2 (1/66) 0.017

First defecation time (d) 3.6 ± 1.8 2.8 ± 1.6 0.007

Diarrhea incidence (%) 46 (31/68) 24 (16/66) 0.012

Urinary catheters time (d) 7.1 ± 2.6 6.6 ± 2.8 0.286

Abdominal cramping (%) 49 (33/68) 23 (15/66) 0.017

Abdominal distension (%) 51 (35/68) 33 (22/66) 0.038

Intake time of fluid diet (d) 3.6 ± 1.2 3.2 ± 1.8 0.131

Intake time of solid diet (d) 5.2 ± 1.6 4.9 ± 1.6 0.280

Side effects of probiotic use 0 0 N/A

Duration of postoperative pyrexia (>38.5°C) (d) 6.98 ± 2.22 6.02 ± 1.68 0.006

Hypoalbuminemia (%) 47 (32/68) 27 (18/66) 0.021

Cumulative duration of antibiotic therapy 7.56 ± 2.26 6.22 ± 1.96 <0.001

Postoperative hospital stay 12.96 ± 3.06 11.26 ± 2.52 <0.001

Hospital charge (Yuan) 58262.36 ± 10262.36 52261.16 ± 12168.28 0.002

Death case 0 0 N/A

Serum zonulin (ng/mg protein) 1.36 ± 0.50 0.42 ± 0.36 <0.001

ALT, alanine transarninase (normal value, 0–40 U/L); AST, aspartate aminotransferase (normal value, 0–40 U/L).

Liu et al. BMC Gastroenterology (2015) 15:34 Page 6 of 13

by the treatment of PRO (ALT, control vs PRO, 56.20 ±18.16 vs 36.28 ± 18.92, P < 0.001; AST, control vs PRO,45.62 ± 22.68 vs 36.18 ± 21.52, P = 0.015, Table 2). Thegastrointestinal recover quicker in the PRO group forPRO group got shorter first defecation time (control vsPRO, 3.6 ± 1.8 vs 2.8 ± 1.6, P = 0.007), less abdominalcramping (control vs PRO, 49% vs 23%, P = 0.017), andabdominal distension (control vs PRO, 51% vs 33%, P =0.038).

PRO reduce the postoperative serum zonulinconcentrationThe serum zonulin was not significantly different betweenthe two groups (0.30 ± 0.62 ng/mg protein vs. 0.36 ±0.38 ng/mg protein, P = 0.502) perioperately. The serumzonulin concentration in the control group (1.36 ±0.50 ng/mg protein) was significantly higher than in thePRO group (0.42 ± 0.36 ng/mg protein, P < 0.001, Table 2)after 10 d postoperative treatment. Results indicated thatPRO lower the postoperative zonulin level efficiently afterCLM surgery. Ccompared with the results in our perviousstudy, which indicated that PRO could reduce the zonulin

level after CRC surgery without liver metastases, the post-operative zonulin level of the control group was higher inour present study of CRC with liver metastases (CLM),compared with that without liver metastases (1.36 ±0.50 ng/mg protein vs 1.08 ± 0.28 ng/mg protein). Andpostoperative septicemia incidence in the control groupwas also higher in our present study (88% vs 73%, P =0.034). Therefore, we hypothesized that another barrieralso played a role in the change of postoperative zonulinlevel inCLM, such as liver barrier. We plan to investigatethe effects of PRO on the liver barrier.We use the intestinal permeability to exclude patients

with intestinal barrier dysfunction. Only patients with nor-mal intestinal permeability at day 3 postoperatively, wereincluded in the following analysis to eliminate the intestinalinterference (Table 3 & Additional file 3: Table S3). We usethe serum zonulin level before treatment to estimate thenormal interval zonulin level. Intestinal permeability wasassessed using the L/M test to distinguish the normal levelof intestinal permeability from higher level at postoperativeday 3. The average L/M ratio (including the ratio of controland PRO groups) before treatment was 0.161 ± 0.059, with

Table 3 Comparison of serum zonulin with the postoperative infectious complications between probiotics and controlthe patients with normal intestinal barrier function (Intention-to-treat)

Outcomes Intention-to-treat

Control (n = 30) PRO (n = 30) P Value

Serum zonulin (ng/mg protein) 0.72 ± 0.26 0.51 ± 0.29 0.004

Septicemia (%)

Total 87 (26/30) 50 (15/30) 0.005

HZ 95 (19/20) 66 (10/15) 0.064

LZ 70 (7/10) 33 (5/15) 0.111

HZ vs. LZ HZ vs LZ (83% vs 48%), P = 0.006

Correlation between septicemia and zonulin, r = 0.613, P < 0.001

Urinary infection (%)

Total 20 (6/30) 0 (0/30) 0.024

HZ 15 (3/20) 0 (0/15) 0.244

LZ 30 (3/10) 0 (0/15) 0.052

HZ vs. LZ HZ vs LZ (9% vs 12%), P = 0.686

Diarrhea incidence (%)

Total 60 (18/30) 27 (8/30) 0.018

HZ 50 (10/20) 27 (4/15) 0.296

LZ 80 (8/10) 27 (4/15) 0.015

HZ vs. LZ HZ vs LZ (40% vs 48%), P = 0.603

Duration of postoperative pyrexia (>38.5°C) (d)

Total 6.99 ± 2.38 5.49 ± 3.21 0.044

HZ 7.97 ± 1.77 (n = 20) 8.02 ± 0.87 (n = 15) 0.927

LZ 5.01 ± 2.27 (n = 10) 2.95 ± 2.61 (n = 15) 0.054

HZ vs. LZ HZ vs LZ (7.99 ± 1.43 vs 3.78 ± 2.64), P < 0.001

Cumulative duration of antibiotic therapy (d)

Total 7.17 ± 1.60 6.13 ± 1.72 0.019

HZ 7.90 ± 1.21 (n = 20) 7.47 ± 1.19 (n = 15) 0.298

LZ 5.70 ± 1.25 (n = 10) 4.80 ± 0.94 (n = 15) 0.052

HZ vs. LZ HZ vs LZ (7.71 ± 1.20 vs 5.16 ± 1.14), P < 0.001

Postoperative hospital stay

Total 12.87 ± 3.01 11.33 ± 2.22 0.029

HZ 14.15 ± 2.64 (n = 20) 13.13 ± 1.60 (n = 15) 0.197

LZ 10.30 ± 1.89 (n = 10) 9.53 ± 0.83 (n = 15) 0.177

HZ vs. LZ HZ vs LZ (13.71 ± 2.28 vs 9.84 ± 1.38), P < 0.001

Hospital charge (Yuan) 60196.12 ± 6532.16 53628.22 ± 6513.28 <0.001

HZ, high serum zonulin (≥0.6 ng/mg protein); LZ, low serum zonulin (<0.6 ng/mg protein); total = HZ + LZ; NS, No significance.Numerical data between groups were compared by the t-test and nominal data by Pearson χ2 test or Fisher’s exact test.HZ vs LZ compares these subgroups without regard to treatment.There was also a significant correlation between zonulin and duration of postoperative pyrexia (r = 0.919, p < 0.001), cumulative duration of antibiotic therapy andzonulin (r = 0.936, p < 0.001), and postoperative hospital stay (r = 0.911, p < 0.001).

Liu et al. BMC Gastroenterology (2015) 15:34 Page 7 of 13

the 95 confidence interval was (0.151, 0.171). Therefore, weonly analyze the patients with the L/M ratio less than 0.171,which were confirmed 3 d after CLM surgery. Further ana-lysis also indicated that there was no significant differencebetween the control and PRO groups before treatment(0.156 ± 0.062 vs 0.167 ± 0.056, P = 0.287).

PRO lowered postoperative infection relatedcomplications in patients with normal intestinal barrierfunctionThe baseline characteristics of patients with normal in-testinal barrier function were re-evaluated. There was nosignificant difference about the baseline characteristics

Figure 2 Probiotics lowered the postoperative bacterialtranslocation and endotoxin (Intention-to-treat analysis). PROeffectively decreased the plasma concentration of endotoxin inpatients of colorectal liver metastases with normal postoperativeintestinal barrier function, compared with the control group (n = 30for control group and n = 30 for PRO group). Black bar representsthe control group, and gray bar represents the PRO group. *(Control) vs. Preoperative, P < 0.05; # (PRO) vs. Preoperative, P < 0.05;* vs. #, P < 0.05. Numerical data are expressed as the means ±standard deviation, and compared by the t-test between groups.Plasma endotoxin was determined preoperatively (hospital admissionday), and postoperatively (10 d treatment after surgery).

Liu et al. BMC Gastroenterology (2015) 15:34 Page 8 of 13

between the two groups (detailed in Additional file 4:Table S4). The postoperative infection-related parame-ters showed significant difference in Table 3 ofintention-to-treat groups were selected for further ana-lysis. For the analysis of patients with normal intestinalbarrier function, serum zonulin level was not signifi-cantly different before treatment (0.32 ± 0.60 ng/mgprotein vs. 0.36 ± 0.39 ng/mg protein, P = 0.761); whilethe serum zonulin concentration after 10 d postopera-tive treatment in the control group was significantlyhigher than in the PRO group (0.72 ± 0.26 ng/mg pro-tein vs. 0.51 ± 0.29 ng/mg protein, P = 0.004, Table 3).The incidence of postoperative septicemia was 87% (26/30) in the control and 50% (15/30) in the PRO group(P = 0.005). Further analysis found that the incidence ofpostoperative septicemia with high serum zonulin levelsdid not differ between the two groups (95 (19/20) vs.66 (10/15), P =0.064). . When patients were groupedaccording to serum zonulin level, the incidence of post-operative septicemia in the high serum zonulin groupwas significantly higher than that in the low serumzonulin group (83% (29/35) vs. 48% (12/25), P =0.006,Table 3). Similar results were indicated in the per-protocol analysis (Additional file 3: Table S3). The dur-ation of postoperative pyrexia, the cumulative durationof antibiotic therapy and postoperative hospital stay allshowed similar results to those of septicemia. The de-tailed comparison between the two groups is shown inTable 3 for the intention-to-treat analysis andAdditional file 3: Table S3 for per-protocol analysis.Furthermore, Spearman’s correlation was used to assessthe relationship between zonulin level and outcome.

PRO got the effective descend on concentration ofplasma endotoxinPatients with normal intestinal barrier function werefurther analyzed. Before treatment, the plasma con-centrations of endotoxin indicated no significant dif-ference (3.32 ± 0.68 in the control group vs. 3.26 ±0.82 in the PRO group, P = 0.759). Control groupshowed a higher level of plasma endotoxin after 10d postoperative treatment (3.96 ± 1.12 vs. 3.32 ± 0.68,P = 0.015), while PRO group showed a lower level ofplasma endotoxin after treatment (2.80 ± 0.88 vs.3.26 ± 0.82, P = 0.041). When the two groups werecompared postoperatively, the PRO group was indi-cated significantly lower level of plasma endotoxin(3.96 ± 1.12 in the control group vs. 2.80 ± 0.88 inthe PRO group, P < 0.001), compared with controlgroup (Figure 2 & Additional file 5: Figure S1).Spearman’s correlation indicated that there was adirect correlation between the postoperative serumzonulin level and the plasma endotoxin (r = 0.962).

PRO reduced postoperative infection rate and positiverate of the blood microbial DNATo determine the types of infectious bacteria and positiverate of microbial culture, bacterial DNA levels in theblood, central line tips and sputum sample cultures wereinvestigated. During the postoperative 72 h period, thegrowth rate of positive bacterial cultures (including blood,central lines and sputum) in the control group (53%, 16 in30 patients) was significantly higher than the PRO group(23%, 7 in 30 patients), as shown in Table 3 (P =0.033).The positive rate of bacteria in the control group (30%, 9in 30 patients) was also found significantly higher com-pared with the PRO group (7%, 2 in 30 patients, P = 0.042,Table 4 & Additional file 6: Table S5). Microbial DNA wasfound in all patients whose blood cultures were positive.The overall blood bacterial DNA positive rate in the con-trol group (53%, 16/30) was significantly higher than inthe PRO group (20%, 6/30, P = 0.015). Only 3 pathogenswere detected in the blood samples, Escherichia coli,Staphylococcus aureus and Aeruginosin; Escherichia coliwas the most common identified bacteria in 47.83% (11/23) of samples with bacteria.

PRO inhibit the p38 MAPK signaling pathwayIt is reported that the p38 MAPK signaling pathway isinvolved in the protection of intestinal barrier function

Table 4 Culture of bacterial culture of blood, central lines and sputum (Intention-to-treat)

Sample Control group (n = 30) PRO group (n = 30)

Bacterium Blood Central lines sputum Blood Central lines sputum

Escherichia coli 6 1 2 1 0 1

Staphylococcus aureus 2 1 2 1 1 1

Klebsiella pneumoniae 0 0 1 0 0 1

Aeruginosin 1 0 0 0 1 0

Bacterial positive patient 9 2 5 2 2 3

Total 16 7

The total bacterial positive rate in control group was 53.33% (16 in 30 patients); in PRO group 23.33% (7 in 30 patients), P = 0.033; bacterial positive rate of theblood in the control group was 30.00% (9 in 30 patients), while in PRO group was 6.67% (2 in 30 patients), P = 0.042.Nominal data by Pearson χ2 test or Fisher’s exact test between groups.

Liu et al. BMC Gastroenterology (2015) 15:34 Page 9 of 13

[27]. In this study, we further investigated the relationbetween p38 MAPK signaling pathway and the liver bar-rier function. To determine whether the PRO treatmentwas associated with the induction of p38 MAPK signal-ing pathway or not, samples from adjacent normal livertissues (>1 cm form the metastatic tumor) were obtainedfrom CLM patients (n = 5 for each group). The resultsshowed that the expression of p38 MAPK was lower inthe PRO group (1.26 ± 0.60), compared with the controlgroup (2.28 ± 0.68, P = 0.033).Randomized clinical trials regarding the effects of peri-

operative PRO treatment on the outcomes of colorectalcancer were rare, especially for CLM. Our perviousstudy indicated the perioperative use of PRO to protecthuman intestinal barrier function and prevent postoper-ative infectious complications after CRC surgery [1,11].No study has reported the effects of pre and postopera-tive use of PRO on the liver barrier after CLM surgery.Here, we first investigated the relation between probio-tics and the liver barrier clinically.In our study, a double-center and double-blind random-

ized clinical trial was performed. Zonulin is a newly identi-fied protein biomarker of barrier function [22,25,28].Interestingly, we found the postoperative zonulin level(1.36 ± 0.50 ng/mg protein) of the control group washigher in CRC with liver metastases compared with thatwithout liver metastases (1.08 ± 0.28 ng/mg protein) [1].So we hypothesize liver barrier also contributed to thechange of postoperative zonulin level in CLM. Resultsshowed that PRO could efficiently lower the postoperativeinfection related complications in CLM patients with nor-mal intestinal barrier function, which indicated that liverbarrier dysfunction also contributed to the postoperativeinfection related complications of CLM patients. A novelconcept of the “clinical regulatory pathway” (CP) wasproposed [1], which was described as the mechanismby which a clinical treatment cause a series of se-quenced molecular and clinical responses and justlooks like the molecular signal transduction pathway.CP could link molecular signals and clinical outcomes

more intuitively and vividly. So we drew a CP pictureto elucidate the relation about PRO, liver barrier, in-testinal barrier and postoperative infection relatedcomplications (Figure 3). We also promote the CP ofthe protective effects of PRO against postoperative in-fection related complications through the liver barrierin CLM patients: peri-operative administration ofPRO might inhibit the p38 MAPK signaling pathway,which will extend the understanding of zonulin aboutits regulation on the barrier function after surgery[29-33].Firstly, the serum zonulin level was not significantly

different between the control and PRO groups, whichwas higher in the PRO group after 10 d postoperativetreatment (Table 2). The incidence of postoperativesepticemia was higher (87%, 26/30) in the control groupcompared with 50% (15/30) in the PRO group. However,either the septicemia rate was not shown significant dif-ference between the two groups, while postoperativesepticemia rate was significantly higher in the highserum zonulin group than the low serum zonulin groupindicated septicemia may be regulated by the organ bar-rier. It shown that incidence of postoperative septicemiawas correlated with the serum zonulin level (r = 0.613for or the intention-to-treat analysis and r = 0.647 forper-protocol analysis). Therefore, we can deduce thatPRO could efficiently reduce the zonulin level and thenlower the incidence of postoperative septicemia (PRO→zonulin→ septicemia).Furthermore, duration of postoperative pyrexia, cumu-

lative duration of antibiotic therapy and postoperativehospital stay were also suggesting the regulation byzonulin. Combined with the, we can deduce that:PRO→ zonulin→ septicemia→ duration of postopera-tive pyrexia time→ duration of antibiotic therapy→postoperative hospital stay→ hospital charge.The finding that Escherichia coli was the main bacteria

identified in blood samples (Table 3) indicated thatpathogen translocated from the intestinal tract to theblood through the liver barrier is the most important

Figure 3 Schematic diagram for the clinical regulatory pathway of probiotics.

Liu et al. BMC Gastroenterology (2015) 15:34 Page 10 of 13

reason for postoperative septicemia in patients withoutintestinal barrier dysfunction [1,11]. We deduce thatperioperative use of PRO may reduce postoperativezonulin levels, injure the liver barrier and so on. (PRO→zonulin→ liver barrier→ septicemia→ duration ofpostoperative pyrexia time→ duration of antibiotic ther-apy→ postoperative hospital stay→ hospital charge).Combined with the overall liver function indexes

(Table 2), both ALT and AST were significantly loweredby the treatment of PRO, indicating the protection ofPRO on liver function, which may contribute to the

injury of liver function [34]. Liver function can be addedto the CP pathway: PRO→ zonulin→ liver function→liver barrier→ septicemia→ duration of postoperativepyrexia time→ duration of antibiotic therapy→ postop-erative hospital stay→ hospital charge.Investigation of the plasma endotoxin indicated no sig-

nificant difference before treatment (P = 0.759). Controlgroup showed a higher level, whereas PRO groupshowed a lower level of plasma endotoxin after 10 dpostoperative treatment, which showed a significantdifference. Spearman’s correlation indicated a direct

Liu et al. BMC Gastroenterology (2015) 15:34 Page 11 of 13

correlation between the postoperative serum zonulinlevel and the plasma endotoxin (r = 0.962). As is well-known that septicemia can be caused by endotoxin[35], we further improve our CP pathway as: PRO→zonulin→ liver function→ liver barrier→ endotoxin→septicemia→ duration of postoperative pyrexia time→duration of antibiotic therapy→ postoperative hospitalstay→ hospital charge.P38 MAPK, a Ser/Thr kinase belonging to the family

of MAPKs, was selected as the signal molecular in theregulation of zonlulin, which is related to the expressionlevel of several inflammatory genes, while inhibition ofp38 MAPK phosphorylation by PRO could protect intes-tinal barrier from dysfunction [27]. Our prevoious studyalso showed that PRO could inhibit the expression ofp38 MAPK, lower the clinical effects of zonulin and de-crease the intestinal permeability [1]. Accordingly, wehypothesize that the effects of PRO were mediated viathe p38 MAPK pathway and then play its role in theliver barrier. Results verified that the expression level ofp38 MAPK was lower in the PRO group compared withthe control group (P = 0.033). Above all, our CP of PROon postoperative septicemia in colorectal cancer surgerycan be presented as: PRO→ p38 MAPK→ zonulin→liver function→ liver barrier→ endotoxin→ septicemia→duration of postoperative pyrexia time→ duration of anti-biotic therapy→ postoperative hospital stay→ hospitalcharge. Combined with the CP that PRO regulate intes-tinal barrier, we deduce that PRO regulate postoperativeinfection related complications in patients of CLM via twopathway—intestinal barrier and liver barrier (Figure 3).It is reported that for severe acute pancreatitis pa-

tients, prophylactic use of PRO could not only fail to re-duce the risk of occurrence of infectious complications,but, on the contrary, increase the patients’ mortality dueto the high oxygen demand and the severe gastrointes-tinal ischemia [36]. While in our study, no death casewas reported, which may because the intestinal barrierand liver barrier was not injured so seriously. One of theadvantages is that although our study is a double→ cen-tered one, the interference of different operators to thestudy results was not the most important [37].We only pointed out one of the possible signal trans-

duction pathways about the zonulin expression regulatedby PRO [38]. It is quite reasonable that zonulin may inturn have a regulation on p38 MAPK pathway or adouble regulation between P38 MAPK and zonulin, givensimilar signaling of the zonulin prokariotic analogue Zot[39],further basic studies may be needed Assessment ofnumerous outcome gives rise to a multiple comparisonissue, especially marginally significant results, which re-quire cautious analyse. Fortunately, the consistency offindings lends support to the effectiveness of treatmentwith PRO [11]. Comparison of HZ and LZ subsets

suffered from small sample size and not randomized com-parison issues and the randomization protection also doesnot apply.

ConclusionTo sum up, perioperative PRO treatment could reducethe rate of postoperative septicemia and maintain theliver barrier in patients undergoing CLM surgery, whichis associated with reduced serum zonulin level. Wepropose a new model about the regulation of PRO toliver barrier via CP (Figure 3). Serum zonulin levelscould be an early biomarker for septicemia. We recom-mend the preoperative oral intake of PRO combinedwith postoperative PRO treatment in patients whounderwent CLM surgery.

Additional files

Additional file 1: Table S1. Baseline of characteristics of the patientswith colorectal liver metastases undergoing surgery at hospital admissionin the study (Per-protocol).

Additional file 2: Table S2. Comparison of postoperative outcomesbetween probiotics and control (Per-protocol).

Additional file 3: Table S3. Comparison of serum zonulin with thepostoperative infectious complications between probiotics and controlthe patients with normal intestinal barrier function (Per-protocol).

Additional file 4: Table S4. Baseline of characteristics of the patientswith normal intestinal barrier function.

Additional file 5: Figure S1. PRO lowered the postoperative bacterialtranslocation and endotoxin (Per-protocol analysis). PRO effectivelydecreased the plasma concentration of endotoxin in patients ofcolorectal liver metastases with normal postoperative intestinal barrierfunction, compared with the control group (n = 28 for control group andn = 29 for PRO group). Black bar represents the control group, andgray bar represents the PRO group. * (Control) vs. Preoperative,P < 0.05; # (PRO) vs. Preoperative, P < 0.05; * vs. #, P < 0.05. Numerical dataare expressed as the means ± standard deviation, and compared by thet-test between groups. Plasma endotoxin was determined preoperatively(hospital admission day), and postoperatively (10 d treatment after surgery).

Additional file 6: Table S5. Culture of bacterial culture of blood,central lines and sputum (Per-protocol).

AbbreviationsCRC: Colorectal cancer; CLM: Colorectal liver metastases; ALT: Alaninetransarninase; AST: Aspartate aminotransferase; Zot: Zonula occludens toxin;BL: Bifido-bacterium longum; BMI: Body mass index; CP: Clinical regulatorypathway; ELISA: Enzyme-linked immunosorbent assay; Hb: Hemoglobin;LA: Lactobacillus acidophilus; LP: Lactobacillus plantarum; L/M: Lactulose/mannitol; IL: Interleukin; PRO: Probiotics; TBS-T: Tris-buffered saline 0.05%Tween 20.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsLZH, QHL and WJP designed the study. LZH wrote the original manuscript,edited all subsequent versions, and performed the study. HMJ, KL and QHLperformed the surgery, and revised the manuscript. KL, ZP, and LCperformed the postoperative abdominal examination. ZP, WL, and PH wereresponsible for data coding and sample analysis. HNQ, LP and PJS performedthe statistical analysis. ZXW, WXJ, and ZP were responsible for assays ofintestinal permeability. LZH, LYX and HMJ were responsible for the molecularbiological determination of p38 MAPK. WXJ, QHL and WJP were responsible

Liu et al. BMC Gastroenterology (2015) 15:34 Page 12 of 13

for the screening and intervention of subjects. All authors approved the finalversion of the manuscript. TC checked the grammer, revised the paper andresponded the reviewers’ questions.

Authors’ informationZhihua Liu, Chao Li, and Meijin Huang are co-first authors.

AcknowledgementsThis work was financially supported by research grants from the NationalNatural Science Foundation of China (No. 81100255 & No. 81370480), andthe New Star of Zhujiang Science and Technology Foundation (No. 2013 J2200023).The clinical trial registration number is ChiCTR→ TRC→ 12002841.

Received: 30 September 2014 Accepted: 27 February 2015

References1. Pedroza-Gonzalez A, Verhoef C, Ijzermans JN, Peppelenbosch MP,

Kwekkeboom J, Verheij J, et al. Activated tumor-infiltrating CD4+ regulatoryT cells restrain antitumor immunity in patients with primary or metastaticliver cancer. Hepatology. 2013;57(1):183–94.

2. Turtoi A, Blomme A, Debois D, Somja J, Delvaux D, Patsos G, et al.Organized proteomic heterogeneity in colorectal cancer liver metastasesand implications for therapies. Hepatology. 2014;59(3):924–34.

3. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancerstatistics. CA Cancer J Clin. 2011;61(2):69–90.

4. Zhao L, Lim SY, Gordon-Weeks AN, Tapmeier TT, Im JH, Cao Y, et al. Recruitmentof a myeloid cell subset (CD11b/Gr1 mid) via CCL2/CCR2 promotes thedevelopment of colorectal cancer liver metastasis. Hepatology.2013;57(2):829–39.

5. Sorbye H, Mauer M, Gruenberger T, Glimelius B, Poston GJ, Schlag PM, et al.Predictive factors for the benefit of perioperative FOLFOX for resectableliver metastasis in colorectal cancer patients (EORTC Intergroup Trial 40983).Ann Surg. 2012;255(3):534–9.

6. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin.2012;62(1):10–29.

7. Shindoh J, Loyer EM, Kopetz S, Boonsirikamchai P, Maru DM, Chun YS, et al.Optimal morphologic response to preoperative chemotherapy: an alternateoutcome end point before resection of hepatic colorectal metastases. J ClinOncol. 2012;30(36):4566–72.

8. Yin Z, Liu C, Chen Y, Bai Y, Shang C, Yin R, et al. Timing of hepatectomy inresectable synchronous colorectal liver metastases (SCRLM): Simultaneous ordelayed? Hepatology. 2013;57(6):2346–57.

9. van Hooft JE, Bemelman WA, Oldenburg B, Marinelli AW, Holzik MF, Grubben MJ,et al. Colonic stenting versus emergency surgery for acute left-sided malignantcolonic obstruction: a multicentre randomised trial. Lancet Oncol.2011;12(4):344–52.

10. Liu ZH, Huang MJ, Zhang XW, Wang L, Huang NQ, Peng H, et al. The effectsof perioperative probiotic treatment on serum zonulin concentration andsubsequent postoperative infectious complications after colorectal cancersurgery: a double-center and double-blind randomized clinical trial. AmJ Clin Nutr. 2013;97(1):1–10.

11. Liu Z, Qin H, Yang Z, Xia Y, Liu W, Yang J, et al. Randomised clinical trial: theeffects of perioperative probiotic treatment on barrier function and post-operativeinfectious complications in colorectal cancer surgery - a double-blind study. AlimentPharmacol Ther. 2011;33(1):50–63.

12. Tran KTC, Smeenk HG, van Eijck CHJ, Kazemier G, Hop WC, Greve JWG, et al.Pylorus preserving pancreaticoduodenectomy versus standard Whippleprocedure - A prospective, randomized, multicenter analysis of 170 patientswith pancreatic and periampullary tumors. Ann Surg. 2004;240(5):738–45.

13. Schroeder RA, Marroquin CE, Bute BP, Khuri S, Henderson WG, Kuo PC.Predictive indices of morbidity and mortality after liver resection. Ann Surg.2006;243(3):373–9.

14. McKenna M. Antibiotic resistance: the last resort. Nature. 2013;499(7459):394–6.

15. Modi SR, Lee HH, Spina CS, Collins JJ. Antibiotic treatment expands theresistance reservoir and ecological network of the phage metagenome.Nature. 2013;499(7457):219–22.

16. Backhed F, Ley RE, Sonnenburg JL, Peterson DA, Gordon JI. Host-bacterialmutualism in the human intestine. Science. 2005;307(5717):1915–20.

17. Liu Z, Ma Y, Yang J, Zhang P, Moyer MP, Qin H. Expression of theLactobacillus Plantarum Surface Layer MIMP Protein Protected NCM460Epithelial Cells from Enteroinvasive Escherichia Coli Infection. Cell PhysiolBiochem. 2011;27(1):99–108.

18. Sugawara G, Nagino M, Nishio H, Ebata T, Takagi K, Asahara T, et al.Perioperative synbiotic treatment to prevent postoperative infectiouscomplications in biliary cancer surgery: a randomized controlled trial. AnnSurg. 2006;244(5):706–14.

19. Rayes N, Seehofer D, Theruvath T, Mogl M, Langrehr JM, Nussler NC, et al.Effect of enteral nutrition and synbiotics on bacterial infection rates afterpylorus-preserving pancreatoduodenectomy: a randomized, double-blindtrial. Ann Surg. 2007;246(1):36–41.

20. Kanazawa H, Nagino M, Kamiya S, Komatsu S, Mayumi T, Takagi K, et al.Synbiotics reduce postoperative infectious complications: a randomizedcontrolled trial in biliary cancer patients undergoing hepatectomy.Langenbecks Arch Surg. 2005;390(2):104–13.

21. Fasano A. Intestinal permeability and its regulation by zonulin: diagnosticand therapeutic implications. Clin Gastroenterol Hepatol. 2012;10(10):1096–100.

22. Tripathi A, Lammers KM, Goldblum S, Shea-Donohue T, Netzel-Arnett S,Buzza MS, et al. Identification of human zonulin, a physiological modulatorof tight junctions, as prehaptoglobin-2. Proc Natl Acad Sci U S A. 2009;106(39):16799–804.

23. Klaus DA, Motal MC, Burger-Klepp U, Marschalek C, Schmidt EM, Lebherz-EichingerD, et al. Increased plasma zonulin in patients with sepsis. Biochem Med (Zagreb).2013;23(1):107–11.

24. Rittirsch D, Flierl MA, Nadeau BA, Day DE, Huber-Lang MS, Grailer JJ, et al.Zonulin as prehaptoglobin2 regulates lung permeability and activates thecomplement system. Am J Physiol Lung Cell Mol Physiol. 2013;304(12):L863–72.

25. Sapone A, de Magistris L, Pietzak M, Clemente MG, Tripathi A, Cucca F, et al.Zonulin upregulation is associated with increased gut permeability insubjects with type 1 diabetes and their relatives. Diabetes.2006;55(5):1443–9.

26. Ghanim H, Sia CL, Upadhyay M, Korzeniewski K, Viswanathan P, AbuayshehS, et al. Orange juice neutralizes the proinflammatory effect of a high-fat,high-carbohydrate meal and prevents endotoxin increase and Toll-likereceptor expression. Am J Clin Nutr. 2010;91(4):940–9.

27. Liu Z, Ma Y, Moyer MP, Zhang P, Shi C, Qin H. Involvement of the MannoseReceptor and p38 Mitogen-Activated Protein Kinase Signaling Pathway ofthe Microdomain of the Integral Membrane Protein after EnteropathogenicEscherichia coli Infection. Infect Immun. 2012;80(4):1343–50.

28. Fasano A, Not T, Wang W, Uzzau S, Berti I, Tommasini A, et al. Zonulin, anewly discovered modulator of intestinal permeability, and its expression incoeliac disease. Lancet. 2000;355(9214):1518–9.

29. Barnes BA, Behringer GE, Wheelock FC, Wilkins EW. Postoperative sepsis:trends and factors influencing sepsis over a 20-year period reviewed in20,000 cases. Ann Surg. 1961;154:585–98.

30. Fried E, Weissman C, Sprung C. Postoperative sepsis. Curr Opin Crit Care.2011;17(4):396–401.

31. Vogel TR, Dombrovskiy VY, Carson JL, Graham AM, Lowry SF. Postoperativesepsis in the United States. Ann Surg. 2010;252(6):1065–71.

32. Litton A, Sinclair RD. Postoperative sepsis in a small surgical unit. Lancet.1963;1(7273):132–4.

33. Wood JA, Colletti E, Mead LE, Ingram D, Porada CD, Zanjani ED, et al.Distinct contribution of human cord blood-derived endothelial colony formingcells to liver and gut in a fetal sheep model. Hepatology. 2012;56(3):1086–96.

34. Ewaschuk J, Endersby R, Thiel D, Diaz H, Backer J, Ma M, et al. Probioticbacteria prevent hepatic damage and maintain colonic barrier function in amouse model of sepsis. Hepatology. 2007;46(3):841–50.

35. Elin RJ, Robinson RA, Levine AS, Wolff SM. Lack of clinical usefulness of thelimulus test in the diagnosis of endotoxemia. N Engl J Med. 1975;293(11):521–4.

36. Besselink MG, van Santvoort HC, Buskens E, Boermeester MA, van Goor H,Timmerman HM, et al. Probiotic prophylaxis in predicted severe acutepancreatitis: a randomised, double-blind, placebo-controlled trial. Lancet.2008;371(9613):651–9.

37. Van Spall HG, Toren A, Kiss A, Fowler RA. Eligibility criteria of randomizedcontrolled trials published in high-impact general medical journals:a systematic sampling review. JAMA. 2007;297(11):1233–40.

38. Topal B, Fieuws S, Aerts R, Weerts J, Feryn T, Roeyen G, et al.Pancreaticojejunostomy versus pancreaticogastrostomy reconstruction after

Liu et al. BMC Gastroenterology (2015) 15:34 Page 13 of 13

pancreaticoduodenectomy for pancreatic or periampullary tumours:a multicentre randomised trial. Lancet Oncol. 2013;14(7):655–62.

39. Goldblum SE, Rai U, Tripathi A, Thakar M, De Leo L, Di Toro N, et al. The activeZot domain (aa 288–293) increases ZO-1 and myosin 1C serine/threoninephosphorylation, alters interaction between ZO-1 and its binding partners,and induces tight junction disassembly through proteinase activated receptor2 activation. FASEB J. 2011;25(1):144–58.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit